Seismic Reflectivity Inversion by Curvelet Deconvolution – A Comparative Study and Further Improvements

Curvelet deconvolution refers to seismic deconvolution for reflectivity inversion based on curvelet transform. Curvelet transform is a multi-scale and multi-directional transform, and thus can provide a sparse representation for seismic reflectivity. When using it to model the reflectivity, the signal is represented effectively by large coefficients and random noise is represented by small ones. In this paper, we conduct a comparative study in the context of reflectivity inversion, to investigate the performance of curvelet deconvolution, the least-squares method and Lp-norm deconvolution. It is shown that by using curvelet deconvolution, the inverted reflectivity profiles have better noise suppression and higher resolution than those obtained by the least-squares method. On the other hand, its results excel those which are obtained by Lp-norm deconvolution in terms of the lateral continuity. Since curvelet deconvolution offers a good trade-off between the lateral continuity and the sparseness, the result obtained by this method can be used as the initial model to enhance the conventional Lp-norm deconvolution. Numerical results show that the lateral continuity of the inversed reflectivity profile has been further improved by the proposed method.